THE ROLE OF COMPOSITE JOINTS ON THE
PERFORMANCE OF STEEL-FRAMED BUILDINGS IN
FIRE
Prof. Khalifa S. Al-Jabri
College of Engineering
Sultan Qaboos University
Oman
World Congress and Exhibition on Construction and Steel Structure 16-18 November, 2015 Dubai. UAE
Beam-to-Column Joints
• Represent a complex structural system that connect
members together
• Various types of Joint exist
• In the design process an engineer must select how to
represent these connections.
• There are multiple ways to represent a connection.
• A pinned connection means that the member (column or
beam) is able to rotate around one axis of that joint or
connection freely while not sliding in any direction.
• A fixed connection is one that doesn’t allow the member
to rotate or slide in any direction.
• Design of joints under normal condition is covered in
EC3-1.8 based on the Component-based approach
Why studying the Behavior of
Joints is important in the event of a
fire?
Beam-to-column joints are one of the structural elements which were found to be
of great significance in enhancing structural behaviour in the event of fire.
They have the ability to re-distribute forces to the adjacent cool regions.
They must maintain structural integrity and the load carrying capacity during
and after fire (i.e. Keep members in contact together)
Their behaviour in fire is different from their behaviour at normal condition
Often difficult to insulate in fire due to their location within structural frame
Their performance in isolation is completely different from their behaviour when
they form part of a structural system due to the structural continuity and restraint to
axial thermal expansion provided by the adjacent connected members in fire
They are a complex structural system
Observations from Real Accidents and
from Full Scale Fire Tests
BEAHVIOUR OF FULL-SCALE STEEL-FRAMED BUILDING
IN FIRE
The Cardington Full-Scale Frame
• Eight–storey steel framed building
• Constructed by the BRE in 1994-95, Cardington UK
• Designed to represent a typical medium rise commercial building
• Plan dimensions are 45x21m, with area of 945m2
1. A retrained beam test
2. A plane frame test
3. The first corner test
4. The second corner test
5. The large compartment test
6. A demonstration test
Cardington Frame Fire Tests
Floor plan of Cardington test frame
showing locations of fire tests
Connection behaviour in real buildings in fire
Vertical
Shear
• Buckling of lower flange of connected
beam.
• Shear buckling in web of connected
beam.
• Large beam deflections (high joint
rotations.
• Some bolt fracture.
Observations from Cardington and other
full-scale tests, and from accidental fires
show:
Ambient temperature:
• Connection subjected
mainly to vertical shear.
Connection behaviour in real buildings in fire
Hogging
Moment
Vertical
Shear
• Buckling of lower flange of connected
beam.
• Shear buckling in web of connected
beam.
• Large beam deflections (high joint
rotations.
• Some bolt fracture.
Observations from Cardington and other
full-scale tests, and from accidental fires
show:
Restrained
expansion
Initial heating stage:
• Beam attempts to
expand – columns and
adjacent structure resist.
Net compression
caused.
• Thermal curvature
generates rotation and
hogging moments.
Initial heating stage:
• Beam attempts to
expand – columns and
adjacent structure resist.
Net compression
caused.
• Thermal curvature
generates rotation and
hogging moments.
Connection behaviour in real buildings in fire
Catenary
Tension
Vertical
Shear
• Buckling of lower flange of connected
beam.
• Shear buckling in web of connected
beam.
• Large beam deflections (high joint
rotations.
• Some bolt fracture.
Observations from Cardington and other
full-scale tests, and from accidental fires
show:
Hogging
Moment
High beam temperature:
• Beam loses strength in
bending – hangs in
catenary. Joints have to
resist catenary tension.
• Large hogging rotation
caused locally by
catenary action.
Tension field
Shear buckling
Axial force in steel downstand of composite beam
400
200
0
-200
-400
-600
-800
0 200 400 600 800 1000 1200
Temperature (°C)
Axial Force (kN)
Axial force in
restrained beam
TENSION
COMPRESSION
Steel strength
Heating
Cooling
Beam flange buckling
Beam shear buckling
End-plate fracture
Joint failures in cooling
1
3
One-sided failures
of partial-depth
end plates
Bolt shear
in fin plate
Nut thread
stripping in
end plate
Fracture in cooling at Cardington
Temperature
Axia
l F
orc
e in
re
str
ain
ed
be
am
COMPRESSION
Cooling
• One-sided failure of partial-depth end
plates during cooling phase.
• Reduced stiffness retains joint integrity.
• Partial fracture may happen when
cooling from net compression …
TENSION
Heating
Partial fracture
Temperature
Axia
l F
orc
e in
re
str
ain
ed
be
am
COMPRESSION
Partial fracture
• One-sided failure of partial-depth end
plates during cooling phase.
• Reduced stiffness retains joint integrity.
• Partial fracture may happen when
cooling from net compression …
• … or tension
TENSION
Heating
Cooling
Fracture in cooling at Cardington
Temperature
Axia
l F
orc
e in
re
str
ain
ed
be
am
COMPRESSION
Partial fracture
• One-sided failure of partial-depth end
plates during cooling phase.
• Reduced stiffness retains joint integrity.
• Partial fracture may happen when
cooling from net compression …
• … or tension
• … or when the net force remains
compressive.
• Shear failure of bolts in fin-plates.
TENSION
Heating
Cooling
Fracture in cooling at Cardington
Outcomes: a few anecdotal observations • Bolts in shear tend to fail by shearing, even at ambient temperature,
although normal design attempts to avoid this.
• Bolts in tension: Thread stripping from nuts can be a problem for single nuts. Closer tolerance nuts/bolts may be needed.
• The rotation at a connection may be very significant in initiating failure at the upper connected element.
• Shear of top bolt in fin plate or web cleat
• Tensile failure/nut stripping of top bolts in other connections
• Tearing of top of partial-depth endplate in shear
• Tearing of top of web cleat in tension
• Shear of top of flush endplate
This may happen either in net tension or net compression.
• The ease with which the lower beam flange can contact the column face may have a considerable effect in reducing the rotation capacity of a connection and initiating above failures.
• Web cleat connections tend to have the best tying resistance, because of excellent rotational capacity and high tying ductility.
• Component-based model of connection zone seems robust enough to include in global modelling.
Local Buckling of Beams
Occurred in most internal beams showed during the heating phase in
the lower flange, and part of the web,
in the vicinity of the joints. Caused by the restraint to thermal
expansion and negative moment
caused by the rotational restraint
from the joint.
The restraint to thermal expansion was provided by the surrounding
cooler structure and the structural
continuity of the test frame. This has taken place in the beam due
to inability of the lower flange of the
beam to transfer the high axial forces
induced in the beam to the adjacent
beams-columns after closure of the
gap in the lower part of the joints
Conservatively, the joints should be assumed as ‘pinned’
and the connected beams as
simply supported allowing larger
mid-span deflections to develop
than when beams are semi-rigidly
connected.
Behaviour of Columns
The internal and external columns were subjected to high moments
causing local squashing of columns. No collapse occurred because the
structure had the ability to carry the
column load using alternative load
path .
These moments are caused by expansion of the connecting beams,
expansion of the heated floor relative
to other floors and induced P-δ
effects during a fire.
Behaviour of The Composite Slab
The composite flooring system used in the Cardington full-scale frame comprised steel downstand beams acting compositely with a floor slab,
constructed using a trapezoidal steel deck, lightweight concrete, and an
anticrack A142 steel mech. The overall depth of the slab was 130mm thick,
with the mesh placed 15mm above the steel deck.
Results from isolated fire tests conducted on beam-to-column composite joints
showed that the composite slab above the joints caused a 20%-30% reduction
in the beam top flange temperatures in comparison with the beam bottom
flange temperature .
This is suggesting that the concrete slab is acting as insulation and as a heat sink to the top of the beam, which enhances joint performance at elevated
temperature.
Observations from the Cardington fire tests and other large building fires have
shown that the behaviour of the composite floor slab plays crucial role in
providing enhanced fire resistance to the structure.
Also it has been confirmed that the performance of a steel frame, with a
composite flooring system, is significantly better than that suggested by the
current fire design methods due to the presence of tensile membrane action in
the composite slab during the fire at large displacements.
Overall Frame Behaviour
• Very large deflections was observed with no
collapse
• Unprotected beams reached temperatures up
to 1100°C
• Restraint to thermal expansion provided by
cold structure had
significant effect on the
survival time of the
building
Tensile Membrane Action
During fire, when significant
numbers of un-protected secondary
steel beams damaged, the lightly
reinforced composite slab acts as a
membrane supported by cold
perimeter beams and protected
columns.
Due to failure of unprotected
steel beams to carry any loading, the
composite slab utilises its full
bending capacity in spanning
between the adjacent cooler
members.
With increasing displacement, the
composite slab acts as a tensile
membrane carrying the loads in the
reinforcement.
In the case of simply supported edges,
the supports will not anchor these tensile
forces and a compressive ring will form
around the edge of the slab.
Failure will only occur at large
displacements with fracture of the
reinforcement.
New Design Method
The beneficial effect of the tensile membrane action at large displacements
that occurred in the composite slab during fire is completely ignored by the
current fire engineering design methods.
The method is valid for both square and rectangular slabs and conforms to
the mode of behaviour observed in the Cardington full-scale fire tests
In order to take advantage of this behaviour when designing structures in fire,
a new design method has been developed for calculating the performance of
steel-framed buildings, with composite flooring systems, subject to fire.
This method uses a simple energy approach to calculate the load-carrying
capacity of a composite flooring system and beams acting compositely.
This method is commonly known as the BRE design method which has
been developed further by Newman into a series of design tables which allow
the designer to leave large numbers of secondary beams unprotected in
buildings requiring 30 and 60 minutes fire resistance although some
compensation features, such as increased mesh size and density may be
required.
• The fact that the frame survived the severe fire conditions raised a number of fundamental issues
whether the current design methods are reflecting the true
behaviour of structures in fire or not.
From the Cardington frame fire tests, the following
observations were drawn
• Unprotected beams in steel deck composite frames can withstand temperatures over 900ºC without any sign of
collapse which indicates that the existing fire design codes
are too conservative, predicting structural collapse at 680ºC.
• Columns are more critical than beams and will need protection in multi-storey buildings
• The behaviour of the structure was different and better than that shown in standard fire tests. The floor slab makes a
major contribution to frame stability and was extremely
beneficial to the survival of the frame in fire
• Local buckling typically occurred in the heated steel beams in the proximity of the connections. Therefore, conservatively,
the connections should be assumed to be pinned.
• Behaviour of the connections during cooling needs to be addressed.
Major areas for future work
• Vertical fire spread makes progressive collapse very
much more likely – design to prevent this makes
structural fire engineering much easier.
• Whole frame/subframe numerical modelling gives
much more complete picture of real behaviour than
prescriptive or isolated member design. But needs to
deal with local failures without treating them as overall
failure. Use dynamic analysis.
• Connection thermo-structural models needed –
especially for ultimate capacity and ductility.
Component based models seem most promising.
Thank you …